Abstract

During the long-term operation process of the satellites in orbit, because of the design life and a variety of environmental factors, the telemetry parameters of satellite will occur slowly and regular changes reflects the decline of its functionality and performance. On the basis of a number patents and patents pending, the telemetry data of satellite firstly can be decomposed into different types of components (i.e. trend, seasonal and random) using x-11 method. Then, it proposed the overall flow of performance prediction. After that, the different prediction methods (such as AR model, BP neural network, non-parametric regression and so on) can be used to predict the performance trends of the telemetry parameters. The experimental telemetry parameters from a satellite power system are used to test the performance and the feasibility of the algorithm, and the results show that the telemetry parameters prediction method based on x-11 method achieves perfect efficiency in predicting the satellite performance.

Keywords: AR, x-11, BP neural network, non-parametric regression, prediction, telemetry parameters.

Graphical Abstract